Wednesday, April 9, 2014

What a 16-Team CFP Playoff Would've Looked Like

The Oklahoma Sooners, your 2013 college football national champions!

Would you have a problem with that?

If you don't, then you must have loved the NCAA basketball tournament, where a fourth-place team from a slightly-less-than-power conference just won the national title. If you do, perhaps you're more of a college football purist who think regular season should matter—a lot.

Either way, we're not here to take sides. Rather, we're here to present some hypotheticals mixed in with facts. Transitive property is used—but not too liberally—to advance a scenario where the Sooners would've won it all last season.

Pundits and critics who disliked the BCS have long advocated for a playoff that involved more than two teams, and they're not even close to being satisfied with the upcoming four-team College Football Playoff. At a minimum, they want 16 teams.

So they'll get 16 teams in our model, and it works because proportionally it best resembles the basketball tournament:

Now, this is how the playoff field at the end of the 2013 regular season would've looked like after the selection committee picked six at-large teams to go with 10 conference champions and then seeded them. The only restriction is that no conference may place more than two at-large entries:

First Round (campus sites)
1. Florida State (ACC) vs. 16. UL-Lafayette (Sun Belt)***
2. Auburn (SEC) vs. 15. Rice (C-USA)***
3. Michigan State (Big Ten) vs. Bowling Green (MAC)***
4. Stanford (Pac-12) vs. Fresno State (MWC)**
5. Baylor (Big 12) vs. Central Florida (AAC)*
6. Alabama (at-large) vs. 11. Oklahoma (at-large)*
7. Ohio State (at-large) vs. 10. Clemson (at-large)*
8. South Carolina (at-large) vs. 9. Oregon (at-large)**

South Carolina just edged Missouri for the last at-large spot from the SEC because it won head-to-head and had a much better out of conference schedule.

Based on results from actual games (*), use of transitive property (**) and simulation (***), these would've been the quarterfinal matchups. We decided to use an NFL-style format where the highest-seeded team always plays the lowest-seeded team instead of using a rigid bracket:

Quarterfinals (campus sites)
1. Florida State vs. 12. Central Florida***
2. Auburn vs. 11 Oklahoma**
3. Michigan State vs. 10. Clemson**
4. Stanford vs. 9. Oregon*

The winning teams then would take a week off before heading to the Rose Bowl and Sugar Bowl for the semifinal games:

1. Florida State vs. 11. Oklahoma (Sugar Bowl)**
3. Michigan State vs. 4. Stanford (Rose Bowl)*

3. Michigan State vs. 11. Oklahoma (AT&T Stadium)**

The Sooners, pulling off a string of upsets thanks to the hot hand of freshman quarterback Trevor Knight, advanced to the national championship game in Arlington ... er, North Texas. In the same JerryWorld where UConn's basketball team completed its improbable run, OU would upstage Michigan State for its own national title.

Is this a just outcome? You decide. Vote in our poll and comment below.


Nick said...

Can you explain the "transitive property" concept a bit? While I'm not saying it couldn't happen, I'm curious as to why you're projecting OU over FSU.

The Guru said...

Yes, we compared common opponents one game removed, i.e., Oregon beat Tennessee and Tennessee beat South Carolina, so we presumed Oregon would beat South Carolina head-to-head.

Of course transitive property isn't accurate in that it doesn't deal with specific matchups, but it's the best we can do absent actual results.

The most difficult transitive property to project is OU over FSU, because FSU didn't lose a game. So we had FSU>Auburn>Alabama by 9 whereas OU>Alabama by 14. It's a stretch, but what would the fun be if FSU won it all, right?

Demosthenes said...

If you have a tournament, you risk a team getting hot and winning it all, even if they may not be the best on paper. If you want the best team on paper to win, don't bother with a playoff, or even a national championship game -- just let the polls handle it.

All things considered, I would prefer a tournament. But four is too few -- this season, though Florida State was the clear #1, there were three or four credible #2 possibilities. Sixteen is too many, though, unless you want to make college football into a semi-pro environment where classes don't matter. I refer you to an eight-team model I recently posted on your blog for what might be a good solution.

The Guru said...

I think four is good, eight has the potential to render regular-season games meaningless. And there are seasons when you really don't need more than two, such as 2005.

In general I oppose having a large playoff field. Most of the time four should get all that's warranted.

Demosthenes said...

I remember people saying that before the BCS, too. "C'mon, most years there are two teams that are clearly better than everyone else!" Those turned out to be words college football fans would come to regret, again and again.

Four will do for now. But eight is the ideal.

The Guru said...

I don't necessarily disagree with that, especially if the committee makes complete hash out of the four-team playoff the next few years.

But in most years, eight would be too many. And unless they're willing to host games on campus, it's not logistically feasible.